Resolved -
The issue is resolved.
A spike in authorization requests exceeded our configured throughput limit. We intentionally throttled this spike due to pre-configured limits, however the requests failed to recover correctly because that failure was cached, which prolonged the impact until those instances were restarted.
We have now mitigated the issue by deploying a follow-up change to ensure startup configuration failures are retried safely rather than persisted.
Mar 10, 14:47 GMT
Investigating -
Users are having issues authorising when signing in. We're investigating the root cause.
Mar 10, 14:37 GMT